Weighting
Weighting survey data is a technique used to bring your sample in line with a specific population. You can do this by giving more or less “weight”, or importance, to specific segments within your sample. The Agile Research weighting feature allows you to adjust the weight of respondents based on demographic or other factors to create more representative survey results in reports.
For example, if you are sending out a survey to an equal number of male and female customers but 55% of survey respondents were male, you could apply the weight feature to align with 50% male and 50% female respondents.
Agile Research uses raking, or Random Iterative Method (RIM) weighting) to weight respondent data to ensure that specific demographics (e.g., gender, age, geography) are represented accurately.
In Agile Research, weighting can be applied to the following question types:
Radio buttons (single select)
Drop down list (single select)
Rating scale
Matrix (single select)
Weighting can also be applied to the following types of metadata:
Respondent metadata
Respondent language
Respondent distribution method
Respondent continent location
Contact metadata
Contact gender
If a survey has very few responses. In general, it is recommended to have at least 30 responses per answer choice.
If the original distribution is very heavily skewed in one direction and you want to change the distribution significantly (e.g., you have 30 female respondents and 1,000 male respondents and want an even distribution of 50% each). In this case, there is not enough data for 30 female participants and, therefore, the weighting would not be meaningful and/or could cause calculation to fail.
Setting up a weight in a survey
To create a new weight variable, complete the following steps:
- A weighting variable calculation will only run when the variable is created and updated. The calculation will not automatically update as new data comes in. If a survey is live and survey data is collected after the weighting variable is created or updated, that new data will not be included in the calculation of the weighting variable. Do not edit questions when they are tied to weighting variables as this could make the data inaccurate or valid.
- Questions that contain answer choices that are populated based on respondents' responses to previous survey questions are not allowed as that can add complexity and inaccuracy to the weighting variable.
In an active report, select the Analyze tab. Then select Weighting.
Select Create new variable.
Enter a name for the variable and, optionally, a description. Select the questions and metadata included in the survey to be used in the weighting variable.
Click Save.
To apply the weighting variable to a report, complete the following steps:
In an active report, select the Analyze tab. Then select Reports.
(Optional) Click Download in the top right of the report and download the raw data as an Excel file.
In the Report Editor, select the Options icon, then Report.
Select the Settings tab. In Data options, select your weighting variable from the Weighting variable dropdown.
The weighting variable that you have applied to a report is shown by the report title.
To edit a variable you have created, select Edit from the Actions dropdown for that variable or select the name of the variable on the Weighting page.
To delete a variable, select Delete from the Actions dropdown for the specific variable.
